The 41-Trial Meta-Analysis: Where the Randomized Evidence Now Sits

A 2026 systematic review pooled 41 randomized controlled trials covering 11,161 men and tested testosterone therapy against the two risks that have shadowed it longest. Neither reached significance: major adverse cardiovascular events, odds ratio 0.83 (95% CI, 0.52–1.32); prostate cancer events, OR 0.88 (0.52–1.51); clinically significant prostate cancer, OR 1.13 (0.39–3.26).[1]

Its own conclusion is measured — current evidence supports the short- to mid-term safety of testosterone therapy, with long-term data still needed.

What it did

• Registration: PROSPERO (CRD42024603054), conducted per PRISMA guidelines.
• Search: PubMed, ClinicalTrials.gov and Cochrane Central. 3,794 records identified; 41 RCTs (n = 11,161) met inclusion criteria.
• Analysis: pooled odds ratios via Mantel-Haenszel or restricted maximum likelihood, under fixed or random-effects models chosen by heterogeneity.
• Additional work: meta-regression to explore heterogeneity and effect modifiers; sensitivity analyses using continuity correction for zero-event trials.

Two design details deserve credit. Prospective registration on PROSPERO commits the analysis plan before results are seen, which is the guard against outcome-shopping. And handling zero-event trials explicitly matters here — for rare outcomes like high-grade prostate cancer, how you treat trials in which nothing happened can move a pooled estimate substantially.

What it found

Outcome	Odds ratio	95% CI	Heterogeneity
Major adverse cardiovascular events	0.83	0.52–1.32	I² = 53.2%
Prostate cancer events	0.88	0.52–1.51	I² = 0.0%
Clinically significant prostate cancer	1.13	0.39–3.26	I² = 0.0%

Comorbidities contributed to the heterogeneity in the MACE analysis — unsurprising, since trials enrolling men with established cardiovascular disease are not measuring the same baseline risk as trials enrolling healthy volunteers.

Reading the confidence intervals honestly

Every one of these intervals comfortably includes 1.0, which is the finding. But they are also wide, and the width is the real information.

The MACE interval runs from 0.52 to 1.32 — consistent with a meaningful reduction and with a modest increase. The clinically significant prostate cancer interval runs from 0.39 to 3.26, which is close to uninformative on its own; it reflects how few such events occur in randomized trials of this size.

So the correct reading is not "testosterone is proven safe." It is that after 41 trials and 11,161 men, no signal of harm has emerged on either endpoint — which, for two risks that carried a regulatory warning for a decade, is a substantive result.

Where it fits with everything else

This analysis is the most recent layer on a stack that now points consistently in one direction:

• Hudson 2022 — individual patient data from 17 trials, 5,601 men, cardiovascular OR 1.07.
• TRAVERSE 2023 — a single adequately powered RCT, 5,246 men, mean 33 months, MACE 7.0% versus 7.3%.
• TRAVERSE prostate 2023 — adjudicated prostate endpoints, no significant difference.
• This meta-analysis, 2026 — 41 RCTs, neither endpoint significant.

Different methods, different populations, different decades of trial data, converging. That convergence is what changed the regulatory position, not any single study.

The limitation the authors keep

Trial durations in this literature are mostly measured in months. TRAVERSE, at a mean 33 months, is the outlier that stretches the window furthest — and testosterone therapy is, for many men, a treatment measured in decades.

"Short- to mid-term safety" is therefore not a hedge added for politeness. It is an accurate description of the evidence base's reach. Nobody has run a randomized, placebo-controlled trial of testosterone over twenty years, and nobody is likely to. The 12-year registry is the closest available substitute, with the design limits that come with it.

Bottom line

Forty-one randomized trials and 11,161 men produce no statistically significant increase in cardiac events, prostate cancer events, or clinically significant prostate cancer on testosterone therapy. The intervals are wide, and the horizon is short to mid-term. Within those limits, this is the current state of the randomized evidence, and it agrees with everything else that has been done properly.

References

1 source

1. García-Becerra CA, Arias-Gallardo MI, Juárez-García JE, et al. Cardiovascular and prostate cancer risk associated to testosterone replacement therapy - a systematic review and meta-analysis of 41 randomized controlled trials. Int J Impot Res. 2026. Published online ahead of print. doi:10.1038/s41443-026-01237-4 · PMID 41673435